WIDENER, Circuit Judge:
Petitioners, National Crushed Stone Association (NCSA), Warren Brothers Company (Warren Brothers), and Arkhola Sand and Gravel Company (Arkhola) seek review of certain regulations promulgated by the Environmental Protection Agency (EPA) pursuant to §§ 301, 304 of the Federal Water Pollution Control Act (FWPCA), 33 U.S.C. §§ 1311, 1314. These regulations establish limitations on the discharge of pollutants1 from existing point sources2 of the crushed stone and construction sand and gravel subcategories of the mineral mining and processing point source category, based upon the best practicable control technology currently available (BPT).3 The regulations challenged here were promulgated in final form on July 12, 1977, to be effective August 11, 1977, 42 F.R. 35843 et seq. Previous to the promulgation of the final regulations, the EPA had issued regulations in “interim final” form,4 June 10,1976, 41 F.R. 23552 et seq. This court has jurisdiction under § 509(b)(1) of the FWPCA, 33 U.S.C. 1369(b)(1).
The crushed stone subcategory regulations, 42 F.R. 35849-50, to be codified as a part of 40 C.F.R., Part 436, subpart B, apply “to the mining or quarrying and the processing of crushed and broken stone and riprap. This subpart includes all types of rock and stone.” 42 F.R. 35849. Riprap consists of large, irregular stones used chiefly in highway embankments and in river and harbor work. Other types of crushed stone are used, for example, in concrete, macadam, and bituminous aggregate, in railroad ballast, in agriculture, and in road bases. Approximately three quarters of all crushed stone is limestone. The crushed stone industry is widespread, with all States reporting some production. The size of individual facilities varies widely, from less than 25,000 to 15 million tons per year. Facilities which produce less than 25,000 tons per year constitute one-third of the total number of facilities, but only 1.3% of total national output. At the other extreme, 5.2% of the facilities each produce more than 900,000 tons annually, but together these make up 39.5% of the total output. Nationwide there are approximately 4800 crushed stone facilities.
[113]*113The construction sand and gravel subcategory regulations, 42 F.R. 35850-51, to be codified as 40 C.F.R., Part 436, subpart C, apply “to the mining and the processing of sand and gravel for construction or fill uses.” 42 F.R. 35850, § 436.30. Construction sand and gravel is used in building, paving, fill and railroad ballast applications. As with crushed stone, sand and gravel facilities are found in all States. Of the more than 5,000 firms engaged in production, 40% have an annual capacity of less than 25,000 tons; these smaller firms account for 4% of the national output. Larger firms with an annual capacity of more than one million tons, on the other hand, account for 12-15% of the national output, although by number they constitute less than 1% of the producing facilities.
Crushed stone and construction sand and gravel operations produce two basic types of waste water which must be discharged and which the EPA has regulated. The first with which the Agency is concerned is that from “mine dewatering.” For crushed stone operations the term means “any water that is impounded or that collects in the mine and is pumped, drained or otherwise removed from the mine through the efforts of the mine operator.” 42 F.R. 35849, § 426.21(b). The definition for the construction sand and gravel industry includes identical language. 42 F.R. 35850, § 436.31(b). The introduction of pollutants includes those coming from “surface runoff of rain water into the mine and mine water treatment systems, ground water seepage and infiltration into the mine.” 42- F.R. 35845. The quantity of mine water that must be discharged either has no correlation with production or is only indirectly related. Only 13% of crushed stone facilities have no mine water. Mine water is also present in construction sand and gravel operations.
The other type of waste water commonly associated with crushed stone and construction sand and gravel operations is that used in the processing of the applicable products. In the crushed stone industry, after the stone has been extracted from the quarry and crushed and screened to meet size specifications, water is added to wash the stone. In a few operations the rough product is processed in a flotation cell, where impurities are removed in the overflow from the cell, and the product is removed in the underflow. Some facilities also have a dry production process. With the dry process, of course, there is no discharge of process generated waste water, although half of the dry process quarries must be dewatered on at least an intermittent basis. Overall in the crushed stone industry 59% of the 4800 facilities wash their product. Of the crushed stone wet processing facilities contacted by the EPA, 33% do not discharge their wash water.
Construction sand and gravel facilities also use water in processing the product to remove impurities such as clay and silt in separating and classifying the product, and in cooling and dust suppression. Half (35) of the facilities visited by the EPA have no discharge of process water because they recirculate all process water. A few facilities achieved no discharge of process water because of soil percolation or because of dredging closed ponds, the process water being discharged back into the pond. Some sand and gravel facilities use a dry process, and. thus have no discharge of process water. 4250 industry plants have wet process operations; only about 750 have dry operations. A few sand and gravel operations use dredging techniques.5
In developing the regulations the EPA considered the varieties, prevalence, and environmental effects of effluent produced by crushed stone and construction sand and gravel operations, and also the current pollution control practices used in the industries. Only two measures of pollution were considered by the Agency to be of sufficient importance to warrant regulation: Total suspended solids (TSS) and pH.6 TSS [114]*114measures both organic and inorganic materials, such as sand, silt, clay, grease, oil, and tar. Solids in suspension interfere with many industrial processes; they are aesthetically displeasing; they burden aquatic life by depleting the oxygen content of water and clogging the respiratory passages of various fauna. The Agency considers TSS to be the single most important pollutant parameter in the mineral mining and processing industry. The petitioners do not challenge the EPA’s regulation of pollution as measured by pH.7
The interim regulations published by the EPA on June 10, 1976, 41 F.R. 23552, divided waste water discharges from crushed stone and construction sand and gravel facilities into two components. “Mine dewa-tering,” referred to earlier, was there defined for both subcategories as “any water that is pumped, drained or otherwise removed from the mine through the direct action of the mine operator.” 41 F.R. 23558, § 436.21(b); 42 F.R. 23559, § 436.-31(b). The definition for construction sand and gravel added “wet pit overflows,” not relevant here. Mine water was permitted to be discharged if TSS concentration did not exceed 30 milligrams per liter (mg/1) of waste water output for any one day. 51 F.R. 23558, § 436.22(a)(2), 41 F.R. 23559, § 436.32(a)(2). The technical material accompanying the crushed stone regulations, 41 F.R. 23554, explained in general that “mine dewatering for all subcategories is limited on a daily maximum basis only, since mine dewatering may occur on an intermittent basis.” Water which collects on quarry floors “is quite clear” and “is typically of excellent purity,” 41 F.R. 23554. Thus, it often may be discharged without treatment, but “in extreme cases [where treatment is necessary] a settling pond at ground level” will permit enough of the suspended solids to settle out so that the mine water will meet the 30 mg/1 criterion. 41 F.R. 23555.
The other waste water discharge regulated by the Agency in the interim regulations was “process generated waste water,” defined for crushed stone operations as “any waste water resulting from the slurry transport of ore or intermediate product, air emissions control, or processing exclusive of mining.” 41 F.R. 23558, § 436.21(e). No discharge of process generated waste water pollutants was permitted by the interim regulations, Id. at § 436.22(a)(1), although the regulations contained exceptions.8 Crushed stone facilities would be able to meet the no discharge requirement by clarifying process generated waste water in a settling pond, and then recirculating it in the production cycle. 41 F.R. 23554. As envisioned by EPA in the interim regulations, all water used in the production processes would be recycled back to the process for reuse, and thus there would be no discharge.
The regulatory scheme established by the interim regulations for construction sand and gravel plants was not identical, but similar. There, too, discharge of process generated waste waters was prohibited. 41 F.R. 23559, § 436.32(a)(1). However, because the EPA found that in construction sand and gravel plants “mine water is often treated in process waste water ponds,” 41 F.R. 23555, the interim regulations for construction sand and gravel operations provided that when “waste streams from various sources are combined for treatment and discharge, the quantity and quality of each pollutant or pollutant property in the combined discharge shall not exceed the quantity and quality of each pollutant or pollutant property allowed had each stream been treated separately.” 41 F.R. 23559, § 436.-32(a)(3). Thus the regulations provided [115]*115that the water to be discharged from the two sources could be commingled in the same settling pond and discharged subject to the 30 mg/1 limit. The regulations did not require or mention recycling, and obviously rejected the Technical Summary, which in 41 F.R. 23555 had recommended recycling where there was a commingling in a settling pond.
On July 12,1977, a little more than a year after the interim regulations were published, the EPA announced its fihal rulemaking for the crushed stone and construction sand and gravel subcategories, 42 F.R. 35843. Two principal changes which concern us were made in the final regulations. First, the maximum TSS mine water effluent limitation for both subcategories was raised from a permissible discharge of 30 mg/1 for any one day to 45 mg/1, but a new thirty day average of 25 mg/1 was promulgated. 42 F.R. 35850, § 436.22(a)(1); 42 F.R. 35851, § 436.32(a)(1). Second, the no process water discharge provision of the interim regulations was changed for some facilities. Instead of the former provision, crushed stone and construction sand and gravel operations “that recycle waste water for use in processing” were permitted to discharge “process generated waste water pollutants” in accordance with a daily maximum of 45 mg/1 and 30 day average of 25 mg/1. 42 F.R. 35850, § 436.22(a)(1), 42 F.R. 35851, § 436.32(a)(1). The interim no discharge provision was retained unchanged for facilities which did not recycle. 42 F.R. 35850, § 436.22(a)(2); 42 F.R. 35851, § 436.32(a)(2). In addition to these changes, the definition of mine water for both subcategories was changed by classifying all water collected or impounded in a mine as “process generated waste water” if the mine is used for treatment of “process generated waste water,”9 and the definition of process generated waste water was amended to “include any other water which becomes commingled with such waste water in a pit, pond, lagoon, mine, or other facility used for treatment of such waste water.”10
In the Summary of Major Changes that accompanied the new regulations, the EPA explained the rationale for the changes described above. The increase in the daily maximum TSS discharge and the addition of a maximum average 30 day discharge “were made because additional data collected since the promulgation of the interim final regulations indicated that the day-today variations in discharges from individual operations were greater than initially found, and because the additional information collected provided the broader data base necessary for formulating a monthly average limitation.” 42 F.R. 35844. The provision permitting discharge of process water for crushed stone and construction sand and gravel facilities that recycle was added when the EPA found that “a number [116]*116of the facilities which currently recycle experience occasional discharges due to natural occurrences, such as rainfall or seepage.” The discharge provision thus was added “to allow a limited discharge of process generated waste water pollutants.” 42 F.R. 35844. Non-recycling facilities were not provided the benefits of the “limited discharge, however, because of the Agency’s view that the best practicable control technology currently available for these industries includes recycling of process water.” 42 F.R. 35844.
Petitioners challenge here the validity of the definition of process generated waste water contained in § 436.21(e) for crushed stone and § 436.31(e) for construction sand and gravel; the TSS limits for process generated waste water and the recycling requirement, § 436.22(a)(1) (crushed stone), § 436.32(a)(1) (construction sand and gravel); the no discharge provision for non-recycling operations, § 436.22(a)(2) (crushed stone), § 436.32(a)(2) (construction sand and gravel), and the TSS limits for mine dewa-tering discharges, § 436.22(a)(3) (crushed stone), § 436.32(a)(3) (construction sand and gravel). In addition, petitioners ask that the variance provisions for the crushed stone (§ 436.22) and construction sand and gravel (§ 436.32) subcategories be set aside as inconsistent with our decision in Appalachian Power Co. v. Train, 545 F.2d 1351 (4th Cir. 1976).
The standards which we must apply to the review of EPA regulations have been set out elsewhere and need not extensively be reviewed here. E. g., Appalachian Power, supra, 545 F.2d at 1356-57; duPont v. Train, 541 F.2d 1018, 1026 (4th Cir. 1976), aff’d in part and rev’d in part on other grounds, 430 U.S. 112, 97 S.Ct. 965, 51 L.Ed.2d 204 (1977); Tanners’ Council of America, Inc. v. Train, 540 F.2d 1188, 1191 (4th Cir. 1976). Briefly, under the Administrative Procedure Act, 5 U.S.C. § 706(2), we may not set aside the regulations unless we find their promulgation to have been “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law,” § 706(2)(A), or “without observance of procedure required by law,” § 706(2)(D). In reviewing these regulations, we are further constrained by “the very basic tenet of administrative law that agencies should be free to fashion their own rules of procedure,” Vermont Yankee Nuclear Power Corp. v. Natural Resources Defense Council, 435 U.S. 519, 544, 98 S.Ct. 1197, 1212, 55 L.Ed.2d 460 (1978), and by provisions for rule making under the Administrative Procedure Act, 5 U.S.C. § 553, which establish “the maximum procedural requirements which Congress was willing to have the courts impose upon agencies in conducting rule making procedures.” Id. at 524, 98 S.Ct. at 1202 (footnote omitted). However, the Agency, as noted, must act in accordance with law, and not in an arbitrary or capricious manner. Neither may it abuse its discretion. 5 U.S.C. § 706. Courts are no longer satisfied with bare administrative ipse dixits, and the Agency must make reasoned decisions with full articulation of the reasoning and take into account all relevant factors. Appalachian Power Company v. EPA, 477 F.2d 495 (4th Cir. 1973).
TSS Effluent Limitations
We first consider the claim of petitioners that the regulations (42 F.R. 35850-1, §§ 436.22(a)(1), 436.22(a)(3), 436.32(a)(1), 436.32(a)(3)) establishing TSS limitations for mine dewatering and process generated waste water are invalid. The EPA has admitted that certain data called the “Versar data” were used to determine the 30 day TSS average of 25 mg/1 for both crushed stone and construction sand and gravel operations, and also has admitted the Versar data were used to determine the TSS daily maximum of 45 mg/1. So, unless the use of the data is harmless, regulations based upon it must be set aside if EPA’s use of the data was not in accordance with law. We think the regulations are invalid for the reasons which follow.
The June 10, 1976 “interim final regulations,” although effective immediately, provided for a public comment period extending until August 9, 1976, 41 F.R. 23553, and NCSA was afforded an additional period in which to comment. NCSA took advantage [117]*117of this opportunity by filing written comments with EPA. After the close of the comment period and a public hearing held on December 2, 1976, however, on December 14-16, 1976 EPA’s contractor, Versar, Inc., visited EPA’s regional headquarters in Atlanta and Dallas and obtained NPDES discharge monitoring reports for various crushed stone operations. This survey, referred to as the Versar data, was completed on February 25, 1977, and on March 15, 1977 the EPA met with some representatives of the crushed stone industry, including NCSA, at which meeting the Versar data were mentioned in the conversation. This was not a public meeting or hearing but was nothing more than a conference held at the request of some members of the industry affected. Petitioners have provided affidavits from participants at the March 15th meeting which show that while the data were discussed at this meeting, they were not made available to the industry for study and analysis. In addition, although industry representatives at the meeting on March 15th requested that they be afforded access to the Versar data, the EPA refused to make the data available until after promulgation of the final regulations. EPA has not filed counter affidavits but has related in its brief that “NCSA was shown the discharge monitoring data compiling the ‘Versar data’ ”. Thus, we are confronted with a fact situation on all fours with that considered by this court in Appalachian Power Co. v. EPA, 579 F.2d 846 (4th Cir. 1978). There, we held that upon an affidavit presented by EPA, petitioners not offering any, in the absence of counter affidavits we would be “unwilling to conclude that the statements in the affidavits are false.” At 853, n. 15. The same result should obtain here, and we thus accept as correct the statements in petitioners’ affidavits that they were not afforded an opportunity to examine, analyze, and comment on the Versar data. The various internal memoranda relied upon by EPA do not contradict the affidavits.
The fact situation so presented to us is very nearly the same as that presented in Portland Cement Assoc. v. Ruckelshaus, 158 U.S.App.D.C. 308, 486 F.2d 375 (1973), in which case the court set aside certain EPA regulations. Prior to the promulgation of the regulations there set aside, the court had remanded to EPA previous regulations because test information upon which the previous regulations had been based was refuted by an engineer experienced in the subject. On remand, instead of commenting on the conclusions of the engineer upon which the remand was based, EPA merely added that analysis to the record. The court set aside the regulations, finding that the comment offered by the industry affidavits was of possible significance in the results of the test. In its discussion of the case, the court said that it found “ . a critical defect in the decision making process in arriving at the standard under review in the initial inability of the petitioners to obtain — in timely fashion — the test results and procedures used in existing plants which formed a partial basis for the emission control level adopted, and in the subsequent seeming refusal of the agency to respond to what seem[s] to be legitimate problems with the methodology of those tests.” 158 U.S.App.D.C. p. 325, 486 F.2d p. 392. The court also stated that “it is not consonant with the purpose of a rule making proceeding to promulgate rules on the basis of inadequate data or on data that [to] critical degree is known only to the agency.” While the second rule just stated was apparently applied by that court in its initial remand proceeding, it is applicable here. EPA admittedly has relied on the Versar data in promulgating the TSS regulations at issue. At the time the regulations were being formulated, only EPA knew about the data in detail. Our case and Portland Cement are no different in that respect. The first rule above mentioned also is applicable to this case. Although the petitioners, or some of them at least, were at the meeting on March 15th, the refusal of the agency to make the actual data available to those immediately affected by it cannot be excused. The. comments of the industry following the promulgation of the interim final regulations and at the December hear[118]*118ing could not have anticipated use of the Versar data because the same had not even been collected by Versar at that time.
We need not, however, place sole reliance on Portland Cement in deciding to remand. We have held in duPont v. Train, 541 F.2d 1018 (4th Cir. 1976), and Appalachian Power Co., 477 F.2d 495 (4th Cir. 1973), that an agency engaged in rule making must “explicate fully its course of inquiry, its analysis and its reasoning.” 541 F.2d at 1026, 477 F.2d at 507. In the case before us, EPA candidly admits that “the development document does not discuss the calculation process by which the agency arrived at the monthly average limit.” That amounts to no less than an admission that the regulation is invalid unless something else appears to render the omission harmless. While EPA, in its brief, does attempt to justify “the path of the administrator’s reasoning,” it has shown us no reason not to apply our holding in duPont that “after the fact rationalization by counsel in brief and argument does not cure non-compliance by the Agency with the stated principles.” 541 F.2d at 1026. See also Portland Cement at p. 395.
The justification offered by EPA in its brief for its failure to give the reasoning behind the new TSS standards is that the 25 mg/1 daily limit was a monthly average from the Versar data, and that the 45 mg/1 daily maximum was an increase sought by industry. The last of the reasons given by EPA is insufficient on its face, for the comments by industry seeking a higher daily maximum discharge limit were in the context of the interim final regulations which had rejected an average discharge limit in favor of a limit for each day. What industry sought was a higher limit for each day, which in fact was effectively lowered on a monthly basis by the new regulation.
More importantly, however, the petitioners never had a chance to respond to the Versar data before the promulgation of the final regulations. This is not consistent with the requirements of law. See Bowman Transportation, Inc. v. Arkansas-Best Freight System, Inc., 419 U.S. 281, 288, n. 4, 95 S.Ct. 438, 42 L.Ed.2d 447 (1974), and Granite City Steel Co. v. EPA, 501 F.2d 925 (7th Cir. 1974).
EPA argues that in all events the use of the Versar data, however, is harmless because the increase in the maximum limit for any one day from 30 to 45 mg/1 more than compensates for the imposition of a monthly maximum average of 25 mg/1, so that, in fact, EPA increased the discharge limits rather than decreased them. The industry contests this conclusion, and EPA calls our attention to no data to corroborate its position.
Following the promulgation of the final regulations, the petitioners did very much the same as the petitioners in Portland Cement. They secured a report by an engineer who was an expert in the field, who took serious issue with the Versar data base on more than one ground. In his opinion the data were insufficient upon which to base the new regulations in many respects, among them: from a practical operating standpoint, the final regulations are more restrictive than the interim final regulations; EPA did not consider the technical feasibility or economic impact of the increase in settling pond size caused by discharging at 25 mg/1 rather than 30 mg/1; seventy-five percent of the permits surveyed were in three States, and over 50 percent in two States, although the industry is scattered nationwide and conditions differ widely; the types of rock mined in the quarries from which the Versar data came was not typical of all of the industry, especially the settling characteristics of the rock from which the Versar data came might well be different from rock in other parts of the country, and as well the settling characteristics in different climates are different; the Versar memorandum does not state the adequacy of its data for statistical analysis; the TSS limitations were arbitrarily selected without benefit of a statistical basis. The objections go on and on, but enough have been related to show that they are far from frivolous, and while EPA will undoubtedly take issue with the report of the engineer, we need not, and do [119]*119not, decide whether the objections raised in the engineer’s report are valid. What we do decide is that the mistakes in the use of the Versar data, if any mistakes there were, were shown by the report to be of possible significance in the formulation of the final regulations. Portland Cement at p. 394. The engineer’s report we have referred to raises significant questions as to the statistical validity of the Versar data, as well as to whether or not the data from that limited base could reliably be used in formulating national regulations. The fact that the petitioners, and just as importantly the fact that the public, had no opportunity to comment on the use of the Versar data prior to the promulgation of the final regulations in the face of serious questions concerning the validity of their use is reason to remand the regulations for further consideration.11 When this is coupled with the fact that the Agency admits that it did not explain the reasons for its actions, the obvious question arises as to whether or not they could have been justified in the record before the Agency.
We remand the TSS limitations to the Agency for reconsideration.12
Recycling Provisions
The next point which we consider is the contention that the recycling provisions (42 F.R. 35850-1, §§ 436.22(a)(1), 436.-32(a)(1)) of the final regulations are invalid. Recycling was not mentioned in any way in the interim final regulations, and there was no requirement in them to recycle. The word simply does not appear. It is true that in the technical summary accompanying the interim final regulations in connection with zero discharge for process generated waste water that recycling was mentioned as an available technology. Also, for construction sand and gravel facilities, recycling was apparently contemplated as a requirement for discharge of process generated waste water when commingled with mine water. Recycling was also referred to as a treatment technology in the economic analysis. But the recycling technology for construction sand and gravel facilities was specifically rejected for the interim final regulations in that the technical summary provided that “treatment other than single settling ponds followed by recycling may be the only economically viable technology.” [120]*120With the technical summary which is at least confusing, the interim final regulations themselves specifically provided, for construction sand and gravel facilities, that process generated waste water, when commingled with other waste water, could be discharged without recycling providing the numbered effluent limitations were met. The case was not quite the same in the interim final regulations for the crushed stone facilities. The interim final regulations contemplated that process generated waste water be kept separate from other waste water, and while providing a zero discharge limit for process generated waste water, allowed discharge of mine dewater-ing water providing the numbered effluent limitations were met.
Comments from the industry on the no discharge provision for process generated waste water, as EPA acknowledges, sought to provide for the crushed stone industry the same right to discharge the commingled waters that EPA had allowed the construction sand and gravel industry in the interim final regulations without recycling.
With this background, the final regulations were promulgated. Petitioners make a multitude of objections to them, of which we will discuss only a few.
First, the term recycle is not defined in the regulations. This may seem somewhat remarkable because the recycling requirement is acknowledged to be a major change between the interim final and the final regulations. This aside, nowhere in the regulations can it be found what part or what volume of the process generated waste water must be recycled to get the benefit of the discharge provisions. If petitioners claim, as EPA claims that they do, that total recycling is required, that term is not defined. If the requirement is partial recycling as EPA claims, then that term is not defined. There simply is no definition.13 EPA claims in its brief that the answer is apparent, that “all the water used in the industrial process must be obtained from the treatment system itself.” But then it adds that “excess water can be treated prior to discharge,” referring to the addition of water from outside the mine to use in the industrial process. If this is a definition, it is not found in the regulations, rather in the brief.
The fact that the regulations do not define recycling may well make them void for vagueness under our decision in duPont, at p. 1033, where we set aside an EPA regulation because we were “not sure what it means in the context in which it is used.” That reasoning might well apply here even if this were the only objection, but there is a more basic fundamental objection to the recycling requirement.
We take it from all the record in this case that in the usual wet operation for both crushed stone and construction sand and gravel that the water used in the industrial process, generally speaking, comes from within the mine or quarry, whether it be from a settling pond or from some other source.14 Assuming only that fact, then, [121]*121the petitioners claim, and EPA points to nothing in the record to refute the fact, that the recycling requirement results in not one bit less discharge of pollutants into the navigable streams than the technology which petitioners claim should be the best practicable, that is allowing settling and discharge without recycling. In its brief, EPA argues that this claim of petitioners is without merit because some water may be lost by evaporation and some may be carried off on the product. But it points to no technical data to support its argument. Without the benefit of an engineering opinion on the point, it would seem to us that if water is used to wash the stone or the sand and the water comes from within the mine, whether from settling pond or otherwise, that the same amount of water is going to be carried out on the finished product regardless of whether it is pumped from the settling pond or from somewhere else. Also, if there is a given amount of water within the mine from whatever source, and evaporation takes place from exposed surface areas, including the industrial process, explanation is required to show why any more water would evaporate in a system utilizing recycling when compared with one which did not. We do not say that petitioners’ position will turn out to be correct, but the record before us does not show it to be incorrect, and it is supported by logic. We decline to accept EPA’s position absent record support.
On the record before us, it is not shown that the addition of the recycling requirement, although admittedly costly and burdensome, will result in any reduction in the discharge of pollutants into the navigable waters. Indeed, EPA as much as acknowledges that absent evaporation and the water carried out on the product petitioners’ premise is correct. This state of facts, then, makes applicable our ruling in du-Pont, at p. 1034, concerning the no discharge provision for plants using the electrolytic process for making hydrogen peroxide. In that case, one plant using a process unique to the industry in this country had a negligible difference in the quality of in-fluent well water and effluent discharge. We noted that the discharges were environmentally insignificant, and importantly we then followed with the statement that “we cannot comprehend how a change from the present to the EPA technology, evaporation and landfill will be beneficial. ... On reconsideration, [for additional reasons] EPA must give consideration to the total environmental impact.”
That holding also applies here. If the requirement of settling plus recycling results in no more benefit to the environment than simply resettling without recycling, then the recycling requirement may not be the best practicable control technology currently available, but only a burdensome and expensive addition. If EPA’s argument is valid, that evaporation and the water carried out on the product result in less pollutants being discharged into the waters, then EPA must at least explain how much less pollutants will be discharged and what the additional cost will be for discharging the lower amount of pollutants. None of these subjects are addressed in the record. See Appalachian Power Co. v. Train, 545 F.2d 1351, 1364 (4th Cir. 1976).
Parenthetically, we note that, especially for the construction sand and gravel industry,15 and to a lesser extent for the crushed stone industry, the problems we have just mentioned concerning recycling could have as easily been avoided as those in the TSS context if only EPA had provided a fair notice of what it was doing with the rea[122]*122sons therefor and made available in time to allow for intelligent comment all of the technical data upon which it acted.
We thus remand the recycling provisions to the Agency for further consideration.
No Discharge Provisions
(42 F.R. 35850-1, §§ 436.22(a)(2), 436.32(a)(2))
The zero discharge requirement for process generated waste water in the interim final regulations (41 F.R. 23558-9, §§ 436.22(a)(1), 436.32(a)(1)) was absolute for the crushed stone industry because those regulations did not contemplate the commingling of process generated waste water with other waste water. It was also absolute for plants in the construction sand and gravel industry which did not commingle process generated waste water with mine dewatering water but permitted discharge without recycling for construction sand and gravel facilities which did so commingle. 41 F.R. 23559, § 436.32(a)(3). In the final regulations, the no discharge requirement for process generated waste water is the same for both industries, that is to say, no discharge for facilities which do not recycle but not applicable to those which do. As we have previously pointed out, recycling was not mentioned in the interim final regulations. The condition for both industries in the final regulations permitting discharge, is recycling, which we have set aside. The condition being set aside, the no discharge provisions should be remanded for reconsideration by the Agency. We do not think the no discharge provisions are meant to stand alone. They were made a part of the final regulations in what EPA contends was an amelioration of the stringency of the interim final regulations in response to industry comments. They can hardly be considered an amelioration if they stand in their original form. Also, on remand, should EPA’s position on the recycling requirement turn out to be not well taken, then it may well decide to omit the recycling requirement which might permit the discharge after settling of commingled process generated waste water and mine water provided numbered effluent limitations are met.
The no discharge provisions are therefore remanded to the Agency for reconsideration.
Variance Provisions
Petitioners also raise the question of whether the crushed stone (42 F.R. 35849, § 436.22) and construction sand and gravel (42 F.R. 35850, § 436.32) variance provisions comport with the decision of this court in Appalachian Power Co. v. Train, 545 F.2d 1351, 1358-60 (4th Cir. 1976), in which case we set aside a variance clause for steam electric power point sources worded substantially the same as the variance provision now under review. Here, as in Appalachian Power, the EPA proposes to grant a variance from the 1977 BPT standards only where “factors relating to the equipment or facilities involved, the process applied, or other such factors relating to such discharges are fundamentally different from the factors considered in the establishment of the guidelines.” Compare 40 C.F.R. § 423.-12(a) (1977) with 42 F.R. 35850, §§ 436.22, 436.32, which show the variance provisions in Appalachian Power and the ones before us to be in the same words.
We held in Appalachian Power that the variance clause was “unduly restrictive,” 545 F.2d at 1359, and ordered a remand to the Agency for the development of “a meaningful variance clause” that would permit economic and other factors to be considered. Id. at 1359-60. The EPA, however, argues that Appalachian Power is not applicable here because review of the variance provision would be premature pri- or to any actual claim for a variance in a discharge permit application. See E. I. du-Pont de Nemours & Co. v. Train, 541 F.2d 1018, 1028 (4th Cir. 1976), aff’d on this ground,, 430 U.S. 112, 128, n. 19, 97 S.Ct. 965, 51 L.Ed.2d 204 (1977). In Appalachian Power, however, we distinguished our decision in duPont, that review of the variance provision would be premature and speculative, because EPA had indicated its refusal in two administrative opinions to consider [123]*123economic factors in considering variance requests. 545 F.2d at 1357-60, n. 22; 39 F.R. 28926-7, dated August 2, 1974, and 39 F.R. 30073 dated August 13, 1974.
We are of opinion the same distinction applies, here, but with added background. On March 3, 1978, the EPA announced its proposed amendments to the steam electric power variance provision we set aside in Appalachian Power. 43 F.R. 8812, 8813. Referring to our decision in Appalachian Power, the EPA announced this change: “In accordance with the Court’s opinion, the [new] variance clause would allow the permit issuers to consider ‘significant cost differentials’ and other economic factors applicable to the particular source involved.” 43 F.R. 8813. But the EPA further stated “[t]his change applies only to steam electric power plants. . . . For categories other than steam electric power plants, economic factors will not be considered in ruling . on BPT variance requests. . . . ” The final steam electric variance amendment appears at 43 F.R. at 43023 et seq., and 44846 (corrected), dated September 22 and 29, 1978, where the same position was taken.
In the meantime, however, despite EPA’s continued assertions that “economic factors” should be “exclude[d]” in all categories of industry regulated by EPA other than steam electric power generation, the administrator, on ruling on a variance in the case of In re Louisiana-Pacific Corp., et al., 10 E.R.C. 1841 (Sept. 15, 1977), had recited that the cost of application of the required technology was a major relevant fact. Despite this recitation, however, EPA did not change its stated position until after this ease was submitted; and the final regulations effecting a change in position are not yet in effect.
On October 17, 1978, the general counsel for EPA withdrew the interpretations of August 1974 upon which we had relied in Appalachian Power in holding that the construction of the variance clause was ripe for review and not premature. 43 F.R. 50042. That withdrawal had been preceded by only a few weeks by the promulgation, on August 21, 1978, of proposed variance provisions which would have made a significant departure from EPA’s previous position and put the Agency more in line with our opinion in Appalachian Power. 43 F.R. 37132. The proposed provisions apparently would apply both to the crushed stone and construction sand and gravel industries, and together with the revocation by EPA of the August 1974 interpretations the matter might be moot were it not for one additional fact. In the paper published in 43 F.R. 50042 withdrawing the August 1974 interpretations, the administrator specifically noted that “EPA continues to believe that § 301(c) of the Clean Water Act (allowing waivers based upon plant-specific economic capability or ‘affordability’) applies only to best available technology (BAT) limitations.”
That construction places EPA squarely in conflict with the rule in Appalachian Power which we have referred to. That case, on pages 1359-1366, specifically required EPA to take into consideration, among other things, the statutory factors set out in § 301(c). EPA is well aware of our position, for, in Appalachian Power, in an order filed September 26, 1977, following the Supreme Court decision in duPont, the same point was made in a motion to us to amend our opinion. The request was denied.
In passing, we should note several arguments of EPA which are not well taken. First, EPA argues that our ruling demands variance requests be based on water quality standards rather than effluent limitations. We rejected that point of view in Appalachian Power Company at page 1378. Second, the administrative action of EPA of October 17, 1978, 43 F.R. 50042, strongly implies that our requirement that § 301(c) factors be at least considered in determining whether or not to grant a variance means that a plant may “secure a BPT variance by alleging [and proving] that the plant’s own financial status is such that it cannot afford to comply with the national BPT limitation.” Like the previous EPA argument, this argument was also specifically rejected by us in Appalachian Power, where we said, [124]*124with reference to the § 301(c) factors, that “if it is doing all that the maximum use of technology within its economic capability will permit and if such use will result in reasonable further progress toward the elimination of the discharge of pollutants ... no reason appears why Consolidated Edison should not be able to secure such a variance should it comply with any other requirements of the variance.” (Emphasis in original)
EPA’s arguments as to water quality standards and as to its interpretation of the consideration of costs under § 301(c) are no better than straw men. Both positions have been previously considered by this court and rejected. They are not even argued by petitioners whose argument here largely is devoted to other specific factors they claim should be considered in determining whether or not to grant a variance.
Finally, we should say that our construction of the variance provisions seems to be generally, if not precisely, in accord with that of the court in Weyerhaeuser Co. v. Costle, 191 U.S.App.D.C. 309, 590 F.2d 1011, 11 E.R.C. 2149 (1978). That court analogized the 1983 variance provisions with the 1977 provisions, drawing upon the Supreme Court opinion in duPont as its authority. In summary, Weyerhaeuser held that a 1977 variance clause must be analogous to the 1983 variance clause, and that EPA’s application of the 1977 variance clause must bear a similar relationship to the 1977 standards as the 1983 variance clause bears to 1983 ^standards. We have held, in Appalachian Power Co., that EPA, in promulgating regulations under the 1977 variance clause, may not exclude the factors to be considered in granting variances under the 1983 standards because the statute contemplates there may be more stringent standards for 1983. While the 1983 standards are not before us for review, we note in the development document that the contractor recommends both pH and TSS limits for 1983 be the same as for 1977. Especially in a case where the effluent limits are the same, but in any case, we think the statute is not meant to stop the operation of a plant in 1979 under 1977 standards under more strict conditions than would apply to a plant operating in 1983 under standards for that year. This situation could easily close a plant in 1979 which would be allowed to operate under a variance in 1983.
Accordingly, we remand the variance provisions to the Agency for compliance with Appalachian Power Company.
Definition of Process Generated Waste Water
Finally, we consider the' petitioner’s contention that the definitions of process generated waste water in the final regulations (42 F.R. 35849-50, §§ 436.21(e), 436.31(e)) are so different from the definitions in the interim final regulations (41 F.R. 23558-9, §§ 436.21(e), 436.31(e)) that they must be set aside.
The addition in the final regulations about which complaint is made is that process generated waste water “shall also include any other water which becomes commingled with such waste water in a pit, pond, lagoon, mine, or other facility used for treatment of such waste water.”
Petitioners view the change as having significantly expanded the definition of process generated waste water, and indeed it has. But we doubt the definition is invalid on its face, for when we take into account the practical consideration that the commingling is done in a settling pond or even in another part of a mine or quarry, and that one, if not the, principal controversy before us seems to be whether effluent discharge of commingled waters meeting numbered effluent limitations is permissible without recycling, then we think the regulation could have a place in the scheme of regulation. This does not take into account, of course, lack of notice. Neither does it take into account the fact that the definition of process generated waste water is completely intertwined with the controversy concerning effluent discharge and the recycling provisions.
Because of this, we think the proper course to take with respect to these regulations is to decline to act on them at this [125]*125time in view of our remand of certain other regulations in this decision. On remand, petitioners should be allowed to comment on the regulations, and if they appear before us again, we will have them in the context of their accompanying regulations and will be in a better position to express an opinion on their validity, both for that reason and because accompanied by comment.
We, therefore, decline to express an opinion upon the validity of the expanded definition of process generated waste water at this time without prejudice to the matter being raised in a subsequent petition following the remand of this case to EPA and its reconsideration of the other regulations we have discussed in this case.
Other Matters Raised in Briefs
The petitioners also have asked us to set aside the regulations for industries which did not have the required technology in place by July 1,1977 on the ground that the July 1, 1977 deadline for compliance with them had passed before the regulations were promulgated on July 12, 1977, to be effective August 11, 1977.
It is apparent that the petitioners’ position raises what may be serious questions of statutory construction as well as constitutional limitations, and because we have remanded the contested regulations on other grounds, we do not express an opinion on those questions.
The fact that we may not have mentioned many of the points raised in the briefs should not infer any opinion of ours as to their merit.
Conclusion
The following regulations are remanded to the Agency for reconsideration:
Crushed Stone Subeategories
Section
§ 436.22 (variance clause)
§ 436.22(a)(1) (TSS limits for process generated waste water and recycling requirement)
§ 436.22(a)(2) (no discharge provision)
§ 436.22(a)(3) (TSS limits for mine dewatering discharge)
Construction Sand and Gravel Subcategories
§ 436.32 (variance clause)
§ 436.32(a)(1) (TSS limits for process generated waste water and recycling requirement)
§ 436.32(a)(2) (no discharge provision)
§ 436.32(a)(3) (TSS limits for mine dewatering discharge)